Did Microsoft's FY26 Q3 disprove the return logic of Azure, Copilot, and AI capital expenditures?

Before the earnings report, my top three concerns about Microsoft were:

First, can Azure continue its high growth?
Second, is Copilot truly being adopted by enterprise customers?
Third, with Microsoft investing such a massive amount in AI capital expenditure, is there any return?

Now that the FY26 Q3 earnings are out, my conclusion is:

Not disproven.

More accurately, the logic behind Azure, Copilot, and AI capex returns has not been disproven; in fact, it continues to be validated.

It's just that Microsoft hasn't yet reached the stage where "all doubts are completely eliminated."

First, look at Azure.

This quarter, Microsoft's total revenue was $82.9 billion, up 18% year-over-year; Microsoft Cloud revenue was $54.5 billion, up 29% year-over-year; Azure and other cloud services revenue grew 40%, or 39% on a constant currency basis. This growth rate is still very strong for a company of Microsoft's size.

More importantly, Microsoft's guidance for Azure next quarter is also good. The company expects Q4 Azure and other cloud services revenue to grow 39%–40% on a constant currency basis, higher than the market's previous expectation of around 36.7%.

So, at least based on this report, the Azure line hasn't shown the market's worst fears of "slowing AI demand" or "a sudden deceleration in cloud growth."

On the contrary, it seems more like:

Demand is still there, and compute and data center delivery capacity remain one of the main constraints.

This is good news for Microsoft itself and a positive signal for the entire AI infrastructure industry chain.

Next, look at Copilot.

The signals from Copilot this time are more substantial than before.

Microsoft 365 Copilot paid seats have reached 20 million, an increase of 5 million from the 15 million disclosed in January. While this number is still not particularly large compared to Microsoft's massive Office user base, adding 5 million paid seats in a quarter at least indicates that the enterprise side isn't completely ignoring it.

Microsoft management also mentioned that the weekly usage frequency for customers using Copilot is already approaching that of Outlook.

I think this point is very important.

The biggest fear for an AI product isn't that no one buys it, but that they buy it and don't use it. If Copilot truly integrates into enterprise workflows rather than remaining in a "trial" phase, its value will gradually shift from a story to actual revenue.

Of course, we need to be objective here: Microsoft has not yet separately disclosed Copilot's revenue, profit margin, renewal rate, or real ROI.

So Copilot is not at the "fully proven" stage, but rather:

It has moved from the storytelling phase into the initial realization phase.

Finally, look at AI capital expenditure.

This is where the market is most conflicted.

Microsoft is indeed spending heavily. FY26 Q3 capital expenditure reached $31.9 billion, up 49% year-over-year; the company expects Q4 capex to be around $40 billion and plans for 2026 capex to be around $190 billion.

Looking at this number alone, it's indeed alarming.

But high capital expenditure itself is not a sin. The real question is: has this money been converted into cloud revenue, AI revenue, and customer usage?

This time, Microsoft disclosed that the annualized revenue run rate for its AI business has reached $37 billion, up 123% year-over-year.

This indicates that Microsoft isn't just burning money on a story; it has started to convert AI infrastructure investments into actual revenue.

Of course, the pressure from AI capex on profit margins and free cash flow still exists, and Microsoft Cloud's gross margin is also affected by AI investments. But at least for now, we can't say "the money was wasted."

Therefore, my assessment of this earnings report is:

Azure is not disproven.
Growth remains strong, and next quarter's guidance is not bad.

Copilot is not disproven.
Paid seats continue to grow, and usage frequency is increasing, but we still need to observe revenue, renewal rates, and real customer ROI.

The logic of AI capex returns is not disproven.
The investment is huge, and the pressure is real, but the AI revenue run rate, Azure growth, and Microsoft Cloud revenue growth all indicate that these investments are yielding returns.

Therefore, Microsoft's earnings report is not the kind that makes you slam the table and shout "across-the-board beat," but it's definitely not a bad report.

It's more like:

An earnings report that continues to validate the AI narrative, but hasn't completely eliminated all market doubts.

For investing, that's actually enough.

My previous thinking remains unchanged:

If Microsoft experiences a significant pullback due to market sentiment, short-term CapEx pressure, or profit-taking, for example, falling back below $400, while the logic behind Azure, Copilot, and AI capex returns is not disproven, then it might be worth considering buying on dips.

But if the stock price remains at high levels, there's no need to chase it.

For a company like Microsoft, the best buying opportunities often aren't when everyone is most euphoric, but when the market temporarily dislikes it for "spending too much, delivering too slowly."

In the short term, it doesn't have the same elasticity as NVIDIA.

But in the medium to long term, Microsoft remains a stable platform company in the AI era.

After this earnings report, my judgment on Microsoft is:

Hold.
Don't chase highs.
If it truly falls to a reasonable range, consider adding to the position.

I believe Microsoft will, like Google last year, use one earnings report after another to prove that AI investment is not just burning money, but can be gradually transformed into sustained revenue and profit growth. So, while the market still debates its AI capex and the stock price hasn't been completely inflated by sentiment, Microsoft is actually a target worth seriously considering for a position.

Personal thoughts, not investment advice.